MAPK4 promotes prostate cancer by concerted activation of androgen and AKT

Tao Shen, … , David D. Moore, Feng Yang

J Clin Invest. 2021;131(4):e135465. https://doi.org/10.1172/JCI135465.

Research Article Oncology

Prostate cancer (PCa) is the second leading cause of cancer death in American men. Androgen receptor (AR) signaling is essential for PCa cell growth/survival and remains a key therapeutic target for lethal castration-resistant PCa (CRPC). GATA2 is a pioneer transcription factor crucial for inducing AR expression/activation. We recently reported that MAPK4, an atypical MAPK, promotes tumor progression via noncanonical activation of AKT. Here, we demonstrated that MAPK4 activated AR by enhancing GATA2 transcriptional expression and stabilizing GATA2 protein through repression of GATA2 ubiquitination/degradation. MAPK4 expression correlated with AR activation in human CRPC. Concerted activation of both GATA2/AR and AKT by MAPK4 promoted PCa cell proliferation, anchorage-independent growth, xenograft growth, and castration resistance. Conversely, knockdown of MAPK4 decreased activation of both AR and AKT and inhibited PCa cell and xenograft growth, including castration-resistant growth. Both GATA2/AR and AKT activation were necessary for MAPK4 tumor-promoting activity. Interestingly, combined overexpression of GATA2 plus a constitutively activated AKT was sufficient to drive PCa growth and castration resistance, shedding light on an alternative, MAPK4-independent tumor-promoting pathway in human PCa. We concluded that MAPK4 promotes PCa growth and castration resistance by cooperating parallel pathways of activating GATA2/AR and AKT and that MAPK4 is a novel therapeutic target in PCa, especially CRPC.

Find the latest version: https://jci.me/135465/pdf The Journal of Clinical Investigation RESEARCH ARTICLE

MAPK4 promotes prostate cancer by concerted activation of androgen receptor and AKT

Tao Shen,1 Wei Wang,1 Wolong Zhou,1 Ilsa Coleman,2 Qinbo Cai,1 Bingning Dong,1 Michael M. Ittmann,3 Chad J. Creighton,4,5 Yingnan Bian,1 Yanling Meng,1,6 David R. Rowley,1 Peter S. Nelson,2,7 David D. Moore,1 and Feng Yang1 1Department of Molecular and Cellular Biology, Baylor College of Medicine, Houston, Texas, USA. 2Fred Hutchinson Cancer Research Center, Seattle, Washington, USA. 3Department of Pathology and Immunology, 4Department of Medicine, and 5Dan L. Duncan Comprehensive Cancer Center, Baylor College of Medicine, Houston, Texas, USA. 6Adrienne Helis Malvin Medical Research Foundation, New Orleans, Louisiana, USA. 7Department of Medicine, University of Washington, Seattle, Washington, USA.

Prostate cancer (PCa) is the second leading cause of cancer death in American men. Androgen receptor (AR) signaling is essential for PCa cell growth/survival and remains a key therapeutic target for lethal castration-resistant PCa (CRPC). GATA2 is a pioneer transcription factor crucial for inducing AR expression/activation. We recently reported that MAPK4, an atypical MAPK, promotes tumor progression via noncanonical activation of AKT. Here, we demonstrated that MAPK4 activated AR by enhancing GATA2 transcriptional expression and stabilizing GATA2 protein through repression of GATA2 ubiquitination/ degradation. MAPK4 expression correlated with AR activation in human CRPC. Concerted activation of both GATA2/AR and AKT by MAPK4 promoted PCa cell proliferation, anchorage-independent growth, xenograft growth, and castration resistance. Conversely, knockdown of MAPK4 decreased activation of both AR and AKT and inhibited PCa cell and xenograft growth, including castration-resistant growth. Both GATA2/AR and AKT activation were necessary for MAPK4 tumor-promoting activity. Interestingly, combined overexpression of GATA2 plus a constitutively activated AKT was sufficient to drive PCa growth and castration resistance, shedding light on an alternative, MAPK4-independent tumor-promoting pathway in human PCa. We concluded that MAPK4 promotes PCa growth and castration resistance by cooperating parallel pathways of activating GATA2/AR and AKT and that MAPK4 is a novel therapeutic target in PCa, especially CRPC.

Introduction cell membrane for activation (14–16). The 2 mTOR complexes, Prostate cancer (PCa) is the most common cancer after skin can- mTORC1 and mTORC2, act coordinately with AKT. mTORC2 is a cer and the second leading cause of cancer death in American major AKT S473 (17), which works together with PDK1, an men (1). Androgen receptor (AR) signaling is essential for PCa AKT T308 kinase (18), to fully activate AKT. In contrast, mTORC1 cell growth and survival. Therefore, androgen deprivation therapy is activated by AKT and is a vital hub that integrates extracellu- (ADT) is the standard therapy for advanced/metastatic PCa (2). lar and nutrient signals to modulate cell growth, autophagy, and However, patients inevitably develop lethal castration-resistant metabolism (19). PTEN, along with PHLPP phosphatases and PCa (CRPC), including resistance to the most advanced thera- INPP4B, offer the major brake to restrain the PI3K/AKT pathway pies: enzalutamide, apalutamide, and abiraterone (3–5). Most (20–23). Aberrant activation of the PI3K/AKT/mTOR signaling CRPCs express activated AR, and attacking AR signaling acti- leads to many pathological outcomes, including cancers (24, 25). vation remains a major therapeutic avenue for CRPC. GATA2 is The PI3K/AKT/mTOR signaling pathway is frequently altered a transcription factor and pioneer factor crucial for inducing AR in PCa (26) and the key components of this pathway have emerged expression and activation in PCa (6–8). GATA2 protein expression as attractive therapeutic targets in PCa and CRPC (27). Complex in human PCa tissues correlates with advanced stages and worse reciprocal crosstalk between the PI3K/AKT/mTOR and AR path- prognosis; it is further elevated in metastatic therapy-resistant PCa ways at multiple levels has been reported. The central theme is (8–11). GATA2 protein is unstable (12, 13) and the molecular mech- mutually antagonistic activities of PI3K/AKT/mTOR and AR (28, anisms that regulate GATA2 stability in PCa remain unknown. 29), although other interactions have also been reported (30–32). AKT plays crucial roles in regulating cell survival, prolif- MAPK4 is an atypical MAPK that lacks the canonical Thr-X-Tyr eration, and metabolism. In the canonical PI3K/AKT/mTOR activation motif for phosphorylation and activation by MAPK kinase pathway, PI3K catalyzes PIP3 production to recruit AKT to the (MAPKK) (33, 34). Little was known about MAPK4 physiological func- tion and its involvement in diseases, including cancers. We recently reported that MAPK4 promotes tumor progression via noncanonical Conflict of interest: The authors have declared that no conflict of interest exists. activation of AKT that is independent of PI3K/PDK1 (35). Here we Copyright: © 2021, American Society for Clinical Investigation. report that MAPK4 also activates AR signaling by both enhancing Submitted: December 3, 2019; Accepted: December 10, 2020; GATA2 mRNA expression and inhibiting GATA2 protein ubiquitina- Published: February 15, 2021. Reference information: J Clin Invest. 2021;131(4):e135465. tion and degradation. Concerted activation of both the GATA2/AR https://doi.org/10.1172/JCI135465. and AKT pathways is required to promote PCa cell growth.

1 RESEARCH ARTICLE The Journal of Clinical Investigation

Results MAPK4 induces androgen-independent activation of AR and MAPK4 promotes PCa cell growth. We recently discovered that greatly enhances the ligand-induced AR activation. To further assess MAPK4 promotes diverse types of human cancers (35). To charac- MAPK4 regulation of AR expression and activation in PCa, we terize MAPK4 activity in androgen-dependent PCa, we first com- measured expression of GATA2, AR/AR-V7, and the AR target pared MAPK4 expression in various human PCa cell lines, includ- PSA and TMPRSS2 in MAPK4-knockdown LAPC4, VCaP, ing VCaP, LAPC4, 22Rv1, LNCaP, C4-2, and DU145 cells, as well and 22Rv1 cells and MAPK4-overexpressing LNCaP, LAPC4, and as PNT1A cells, an immortalized human prostate epithelial cell line. 22Rv1 cells. In line with MAPK4 regulation of GATA2 and AR/ MAPK4 is expressed at high levels in VCaP, LAPC4, and 22Rv1 cells, AR-V7 protein expression, MAPK4 promoted expression of the AR with low expression in LNCaP, C4-2, DU145, and PNT1A cells (Fig- targets in all of the cell lines (Figure 3, D–F and Figure 4, D–F). ure 1A). To assess MAPK4 function in PCa, we performed knock- MAPK4 induction also both promoted basal, androgen- down of MAPK4 in VCaP, LAPC4, and 22Rv1 cells (see below). As independent AR activation and supported androgen-depen- expected, MAPK4 knockdown greatly inhibited LAPC4, VCaP, and dent AR activation. Dox-induced MAPK4 overexpression sig- 22Rv1 cell proliferation (Figure 1, B and C) as well as anchorage- nificantly activated both basal AR activity in LNCaP and LAPC4 independent growth (Figure 1D). Thus, MAPK4 plays a critical role cells cultured in 10% charcoal-stripped serum (CSS) plus 10 μM in promoting the growth of MAPK4-high PCa cells. MDV3100 for maximum androgen ligand blockade, as well as AR We next engineered ectopic MAPK4 overexpression in LNCaP, agonist R1881-induced AR activation of PSA mRNA expression LAPC4, and 22Rv1 cells (see below). In general, MAPK4 overex- (Figure 4, G and H). In accord with this, knockdown of MAPK4 pression promoted LNCaP, LAPC4, and 22Rv1 cell proliferation greatly inhibited both baseline and the ligand-stimulated AR acti- as well as anchorage-independent growth (Figure 2, A–G). More vation in LAPC4 cells (Figure 3G). specifically, proliferation and anchorage-independent growth of MAPK4-induced AR activation crucially supports the growth the androgen-dependent control LNCaP and LAPC4 cells were promoting activity of MAPK4 in PCa. To assess the functional sig- minimal when cultured in complete androgen blockade conditions nificance of AR signaling in mediating MAPK4 tumor-promoting consisting of 5% charcoal-striped serum (CSS) deprived of andro- activity in PCa, we used AR knockdown in the MAPK4-overex- gen plus cotreatment of 10 μM MDV3100. However, MAPK4 over- pressing LNCaP cells. As expected from the critical roles of AR expression potently stimulated LNCaP and LAPC4 cell growth, in PCa, AR knockdown significantly repressed MAPK4-induced including anchorage-independent growth under such conditions LNCaP cell growth (Figure 5, A and B; Supplemental Figure 1A; (Figure 2, A, B, E, and F). MAPK4 overexpression also robustly pro- supplemental material available online with this article; https:// moted PNT1A cell growth and transformed these cells into anchor- doi.org/10.1172/JCI135465DS1). In accord with this, AR overex- age-independent growth, as we previously reported (35). Alto- pression in control LAPC4 cells minimally affected cell growth, but gether, these results demonstrate that MAPK4 promotes growth, largely rescued MAPK4-knockdown LAPC4 cell growth (Figure 5, including androgen-independent growth of PCa cells in vitro. C and D) and AR target expression (Supplemental Figure 1B). MAPK4 induces GATA2 and AR expression in PCa. Since AR Altogether, these data suggest that MAPK4-induced AR signaling plays an essential role in driving most PCa and GATA2 is a key crucially supports MAPK4 growth promoting activity in PCa. transcription factor regulating AR expression and activation, we MAPK4 enhances GATA2 to induce AR expression and activation examined the expression of MAPK4, AR, and GATA2 in human and to promote PCa cell growth. Since GATA2 is essential for AR PCa cells (Figure 1A). Expression of AR and GATA2 generally expression and activation in PCa (7), we next investigated GATA2 followed that of MAPK4 in these cells, suggesting that MAPK4 function in mediating MAPK4-induced AR signaling and PCa may regulate GATA2 and AR expression. Indeed, knockdown of cell growth. GATA2 knockdown in LNCaP-iMAPK4 cells greatly MAPK4 in the LAPC4 cells using 5 independent shRNAs greatly inhibited MAPK4-induced AR expression and PSA expression, repressed GATA2 and AR protein expression (Figure 3A). Knock- and cell growth, supporting the critical roles of GATA2 in medi- down of MAPK4 in VCaP cells and Dox-induced knockdown ating MAPK4 biology in these cells (Figure 5, E and F; Supple- of MAPK4 in LAPC4 cells similarly suppressed GATA2 and AR mental Figure 1A). Furthermore, GATA2 knockdown in control expression (Figure 3, A and B). In the opposite direction, consti- LAPC4 cells reduced AR expression and cell growth to levels tutive overexpression of MAPK4 in LNCaP cells as well as Dox- similar to those of MAPK4-knockdown cells, and GATA2 knock- induced overexpression of MAPK4 in LNCaP cells and LAPC4 down in MAPK4-knockdown LAPC4 cells further reduced cell cells all greatly increased GATA2 and AR protein levels (Fig- growth (Supplemental Figure 1C). We made similar observations ure 4, A and B). AR-V7 is an AR splicing variant with androgen- in VCaP cells (Supplemental Figure 1D). Therefore, knockdown independent constitutive activity, and gain of AR-V7 expression of GATA2 in MAPK4-high PCa cells at least partially recapitulates is an established mechanism for castration-resistant growth in MAPK4-knockdown phenotype. CRPC (36–41). 22Rv-1 cells highly express AR-V7 as well as full- To further define the functional significance of GATA2 in length AR at lower level (38, 39). Interestingly, Dox-inducible mediating MAPK4 activity in PCa, we also performed gain-of- knockdown of MAPK4 expression in 22Rv1 cells repressed expres- function studies. Ectopic expression of GATA2, like MAPK4, sion of GATA2 and AR-V7, as well as full-length AR (Figure 3C). In largely rescued AR expression and activation as well as the growth, accord with this, Dox-inducible ectopic overexpression of MAPK4 including anchorage-independent growth of MAPK4-knockdown enhanced expression of GATA2, AR-V7, and full-length AR in LAPC4 cells (Figure 5, G–J). Altogether, these data suggest that 22Rv1 cells (Figure 4C). Altogether, these data strongly support MAPK4-induced GATA2/AR signaling crucially supports MAPK4 that MAPK4 promotes GATA2 and AR/AR-V7 expression in PCa. growth promoting activity in PCa.

2 J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 The Journal of Clinical Investigation RESEARCH ARTICLE

Figure 1. Knockdown of MAPK4 inhibits PCa cell growth. (A) MAPK4, GATA2, and AR expression in human PCa cell lines as well as in PNT1A, an immortal- ized normal prostate epithelial cell line that does not express AR. Left panel: Western blots. Right panel: qPCR. (B) MTT assays comparing the growth of the LAPC4 cells with Dox-inducible knockdown of MAPK4 (iG2, iG4) or control (iNT) and VCaP cells with knockdown of MAPK4 (G2, G4) or control (NT). Also shown are 22Rv1 cells with Dox-inducible knockdown of MAPK4 (iG2, iG4) or control (iNT) cultured under complete castration (CAS) condition in media con- taining 5% charcoal-stripped serum (CSS) plus 10 μM MDV3100 for maximal androgen blockade. Data represent mean ± SD. (C) BrdU incorporation assays comparing the proliferation of 22Rv1 cells with Dox-inducible knockdown of MAPK4 (iG2, iG4) or control (iNT) cultured under CAS condition as described above. Original magnification: ×400. The percentage of BrdU-labeled cells was quantified and data shown as mean ± SEM. (D) Soft-agar assays compar- ing the anchorage-independent growth of the LAPC4 cells with Dox-inducible knockdown of MAPK4 (iG2) or control (iNT) and VCaP cells with knockdown of MAPK4 (G2) or control (NT). Also shown are 22Rv1 cells with Dox-inducible knockdown of MAPK4 (iG2) or control (iNT) cultured under CAS condition as described above. Original magnification: ×50. The colony numbers were quantified and data shown as mean ± SEM. P values determined by unpaired 2-tailed Student’s t test and adjusted P values determined by 1-way ANOVA followed by Dunnett’s multiple comparisons (B, C) or 2-way ANOVA followed by Sidak’s multiple comparisons (LAPC4 data in D). ****P ≤ 0.0001. Data are representative of at least 3 independent experiments.

J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 3 RESEARCH ARTICLE The Journal of Clinical Investigation

Figure 2. MAPK4 overexpression promotes PCa cell growth. (A–C) MTT assays comparing the growth of LNCaP (A), LAPC4 (B), and 22Rv1 (C) cells with Dox-inducible overexpression of MAPK4 (iMAPK4) or control (iCtrl). FBS: cells cultured in media containing 10% FBS. CAS: cells cultured under CAS condi- tion in media containing 5% CSS plus 10 μM MDV3100. Data represent mean ± SD. (D) BrdU incorporation assays comparing the proliferation of LNCaP cells with Dox-inducible expression of MAPK4 (iMAPK4) or control (iCtrl). Original magnification: ×400. The percentage of BrdU-labeled cells were quantified and data shown as mean ± SEM. (E–G) Soft-agar assays comparing the anchorage-independent growth of LNCaP (E), LAPC4 (F), and 22Rv1 (G) cells with Dox-inducible overexpression of MAPK4 (iMAPK4) or control (iCtrl). FBS and CAS: cell culture conditions as described above. Original magnification: ×50. The colony numbers were quantified and data shown as mean ± SEM. P values determined by unpaired 2-tailed Student’s t test. ***P ≤ 0.001. ****P ≤ 0.0001. Data are representative of at least 3 independent experiments.

MAPK4 promotes GATA2 expression by both inducing GATA2 MAPK4 also regulates GATA2 protein stability, we first investigat- gene transcription and enhancing GATA2 protein stability. MAPK4 ed MAPK4 regulation of GATA2 protein levels when GATA2 mRNA promoted GATA2 gene transcription (Figure 3, D–F and Figure 4, expression levels were comparable. Accordingly, we transfected D–F). However, GATA2 protein is unstable and subject to ubiqui- HA-tagged GATA2 into HEK293T cells together with increasing tin-proteasome mediated degradation (12, 13). To assess whether doses of a Flag/His-tagged MAPK4. As expected, increasing doses

4 J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 The Journal of Clinical Investigation RESEARCH ARTICLE

Figure 3. Knockdown of MAPK4 represses GATA2/AR expression and AR activation in PCa cells. Western blots on (A) LAPC4 and VCaP cells with knock- down of MAPK4 (G2, G4, G5, G6, G7) or control (NT), (B) LAPC4 cells with Dox-inducible knockdown of MAPK4 (iG2) or control (iNT), and (C) 22Rv1 cells with Dox-inducible knockdown of MAPK4 (iG2 and iG4) or control (iNT). qPCR on (D) LAPC4, (E) VCaP, and (F) 22Rv1 cells with knockdown of MAPK4 (G2, G4) or control (NT), or Dox-inducible knockdown of MAPK4 (iG2, iG4) or control (iNT). (G) LAPC4 cells with Dox-inducible knockdown of MAPK4 (iG2, iG4) or con- trol (iNT) were treated with 10 μM MDV3100, Ctrl, or increasing doses (0.01, 0.1, and 1 nM) of R1881 in 10% CSS in the presence of 2 μg/mL Dox. qPCR was used for assessing PSA expression. Data represent mean ± SD. Adjusted P values determined by 1-way (D–F) or 2-way ANOVA (G) followed by Dunnett’s multiple comparisons. **P ≤ 0.01. ***P ≤ 0.001. ****P ≤ 0.0001. Data are representative of at least 3 independent experiments.

of MAPK4 exhibited limited effect on mRNA levels of the ectop- We next examined how MAPK4 regulates endogenous ically expressed GATA2. In sharp contrast, MAPK4 substantial- GATA2 protein stability in PCa cells. While Dox-induced ly enhanced the protein levels of this ectopically overexpressed MAPK4 expression in the LNCaP-iMAPK4 cells induced GATA2 GATA2 in a dose-dependent manner (Figure 6A). This suggests gene transcription (Figure 4D), blocking protein synthesis using that MAPK4 is capable of potently regulating GATA2 protein levels. cycloheximide (CHX) revealed that MAPK4 also increased

J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 5 RESEARCH ARTICLE The Journal of Clinical Investigation

Figure 4. MAPK4 induces GATA2/AR expression and AR activation in PCa cells. Western blots on (A) LNCaP cells with overexpression of MAPK4, control (Ctrl), Dox-inducible overexpression of MAPK4 (iMAPK4) or control (iCtrl), (B) LAPC4 cells, and (C) 22Rv1 cells with Dox-inducible overexpression of MAPK4 (iMAPK4) or control (iCtrl). Additional loading control (β-ACTIN) was used for the ectopic MAPK4 expression in 22Rv1-iMAPK4 cells. The left and right pan- els of Western blots were on the same samples run at different times. qPCR on (D) LNCaP, (E) LAPC4, and (F) 22Rv1 cells with Dox-inducible overexpres- sion of MAPK4 (iMAPK4) or control (iCtrl). (G) LNCaP-iMAPK4 cells and (H) LAPC4-iMAPK4 cells were treated with 10 μM MDV3100, vehicle control (Ctrl), or increasing doses (up to 10 nM) of R1881 in 10% of CSS in the presence (+) or absence (–) of 0.5 μg/mL Dox. qPCR was used for assessing PSA expression. Data represent mean ± SD. P values determined by unpaired 2-tailed Student’s t test. ***P ≤ 0.001. ****P ≤ 0.0001. Data are representative of at least 3 independent experiments.

GATA2 protein stability (Figure 6B). While endogenous and that MAPK4 promotes GATA2 expression through both enhanc- ectopically overexpressed GATA2 protein were relatively sta- ing GATA2 gene transcription (Figure 3D and Figure 4E) and ble in CHX-treated LAPC4 cells, MAPK4 knockdown greatly inhibiting GATA2 protein degradation. Ubiquitination assays reduced their stability (Figure 6, C and D), and proteasome revealed that MAPK4 greatly reduced GATA2 ubiquitination in inhibitor MG132 treatment partially blocked this reduction (Fig- both LNCaP cells and HEK293T cells, further confirming that ure 6E). This partial rescue of GATA2 levels in the MG132-treat- MAPK4 enhances GATA2 protein stability by repressing GATA2 ed MAPK4-knockdown LAPC4 cells is consistent with the notion protein ubiquitination and degradation (Figure 6F).

6 J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 The Journal of Clinical Investigation RESEARCH ARTICLE

Figure 5. GATA2/AR signaling is essential for mediating MAPK4 tumor-promoting activity in PCa. (A) Western blots and (B) Proliferation assays on the LNCaP cells with Dox-induced MAPK4 overexpression (iMAPK4) or control (iCtrl) transfected with siRNA against AR (siAR-1, siAR-2) or luciferase (siLUC, negative control). (C) Western blots and (D) Proliferation assays on the LAPC4 cells with Dox-induced MAPK4 knockdown (iG4) that also ectopically express AR or Control (Ctrl), or control (iNT). (E) Western blots and (F) Proliferation assays on the Dox-induced LNCaP-iMAPK4 or LNCaP-iCtrl cells trans- fected with siRNA against GATA2 (siGATA2-1, siGATA2-2) or luciferase (siLUC). (G) Western blots, (H) qPCR analysis, (I) proliferation assays, and (J) soft- agar assays (original magnification: ×50) on the Dox-induced MAPK4-knockdown LAPC4 (iKD) cells that also overexpress MAPK4, GATA2, or control (Ctrl). LAPC4-iNT: Dox-induced nontargeting control LAPC4 cells. In H, data represent mean ± SD. All other data represent mean ± SEM. P values determined by unpaired 2-tailed Student’s t test and adjusted P values determined by 1-way ANOVA followed by Sidak’s multiple comparisons. ****P ≤ 0.0001. Data are representative of at least 3 independent experiments.

J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 7 RESEARCH ARTICLE The Journal of Clinical Investigation

Figure 6. MAPK4 enhances GATA2 protein levels in PCa. (A) HEK293T cells were cotransfected with vectors carrying HA-tagged GATA2 (HA-GATA2) and increasing doses of Flag/His-tagged MAPK4 (MAPK4-FH). Left panel: Western blots. Right panel: qPCR. Western blots were performed using cell lysates from (B) LNCaP-iMAPK4 (Dox-inducible MAPK4 expression) cells induced with 0.5 μg/mL Dox, (C) LAPC4-iNT (control) and LAPC4-iG2 (Dox-inducible knockdown of MAPK4) cells induced with 2 μg/mL Dox, (D) LAPC4-NT (control) and LAPC4-G2 (stable knockdown of MAPK4) cells with 0.5 μg/mL Dox- induced ectopic expression of GATA2 (iGATA2). All cells were also treated with (B) 3 μg/mL or (C, D) 10 μg/mL CHX for up to 4 hours as indicated. (E) Western blots on LAPC4 cells with stable knockdown of MAPK4 (G2, G4) or control (NT). All cells were also treated with 10 μM MG132 for 4 hours. (F) Ubiquitination assay. LNCaP cells and HEK293T cells were transfected with plasmids expressing Myc-tagged GATA2 (Myc-GATA2), Flag/His-tagged MAPK4 (MAPK4-FH), HA-tagged Ubiquitin (HA-Ubi), and/or vector controls. Twenty-four hours later, cells were treated with 10 μM MG132 for 6 hours before collection. Immunoprecipitation were performed using anti-Myc tag antibody, and Western blots were run using anti-HA antibody. WCL: whole cell lysate. Coimmunoprecipitation (co-IP) assays reveal (G) ectopically overexpressed Myc-GATA2 binding to Flag/His-tagged MAPK4 and MAPK4S186A mutant (MAPK4-HF and S186-HF) in the transfected HEK293T cells, and (H) endogenous GATA2 binding to endogenous MAPK4 in VCaP cells. (I) BiFC assay for GATA2 association with MAPK4 in cytoplasm. HeLa cells were cotransfected with YN-MAPK4 and YC-GATA2 or YC control vectors. Twenty-four hours later, the cells were fixed, counterstained with DAPI, and imaged for YFP fluorescence to indicate MAPK4-GATA2 interaction. Original magnification: ×400. Data are representative of at least 3 independent experiments.

8 J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 The Journal of Clinical Investigation RESEARCH ARTICLE

To assess whether MAPK4 directly regulates GATA2 pro- growth-promoting activity in LNCaP-iMAPK4D254A cells (Figure tein ubiquitination and degradation, we further investigated 7G) and in the stably engineered LNCaP cell lines (Figure 7H). MAPK4-GATA2 interaction. When ectopically overexpressed These results further support the essential role of AKT activation in HEK293T cells, a Myc-tagged GATA2 coimmunoprecipitated in mediating MAPK4-dependent PCa cell growth. (co-IP) with a Flag-tagged MAPK4 (Figure 6G). We also detected Concerted activation of AKT and GATA2/AR signaling is suf- the co-IP of endogenous MAPK4 and endogenous GATA2 in VCaP ficient to promote PCa growth and castration resistance. Since AKT cells (Figure 6H). Bimolecular fluorescence complementation and GATA2/AR signaling pathways appeared to work concertedly (BiFC) enables visualization of protein-protein interactions and downstream of MAPK4, we next investigated whether simulta- subcellular locations in live cells (42). Accordingly, we co-trans- neous activation of AKT (expression of AKT1[DD]) and GATA2 fected HeLa cells with a YN-MAPK4 construct encoding MAPK4 overexpression in the absence of ectopic MAPK4 overexpression fused to the N-terminal fragment of YFP, along with a YC-GATA2 promotes PCa growth and castration resistance. While expression construct encoding GATA2 fused to the C-terminal fragment of of AKT1(DD) or GATA2 alone exhibited some activities, the coex- YFP. Observation of a positive cytoplasmic signal in cells coex- pression of both substantially enhanced LNCaP cell growth (Figure pressing YN-MAPK4 and YC-GATA2, but not in the YN-MAPK4 7H). Furthermore, while GATA2 growth-promoting activity was or YC vector control cotransfected cells, further confirmed the markedly reduced in engineered LNCaP cells cultured in 5% CSS MAPK4-GATA2 interaction and revealed that this interaction deprived of androgen, coexpression of AKT1(DD) and GATA2 was occurs mainly in the cytoplasm (Figure 6I). sufficient to induce robust proliferation under the same condition Altogether, these data suggest that besides promoting GATA2 (Figure 7H). These data support that the concerted activation of gene transcription, MAPK4 binds GATA2 and further enhances AKT and GATA2/AR signaling is sufficient to promote PCa growth GATA2 protein levels by stabilizing GATA2 protein through inhib- and castration resistance. Finally, ectopic overexpression of GATA2 iting its proteasome-mediated degradation. or AKT1(DD) alone was sufficient to mostly rescue the growth of AKT activation is essential for mediating the growth-promoting the MAPK4-knockdown VCaP cells (Figure 7I). These data suggest activity of MAPK4 in PCa. We previously reported that MAPK4 pro- that the remaining MAPK4 activity in the MAPK4-knockdown motes tumor growth via noncanonically phosphorylating and acti- VCaP cells may be sufficient to support VCaP cell growth once sup- vating AKT (35). As expected, MAPK4 knockdown also repressed plemented with either GATA2 or AKT1 activation. AKT phosphorylation in PCa cells (Figure 7A). The mutation of MAPK4 promotes PCa xenograft growth and castration resis- MAPK4 residues D254A or S186A largely abolished MAPK4 acti- tance, and its expression correlates with AR activation in human vation of AKT via different mechanisms (35). Interestingly, when CRPC tissues. We further investigated MAPK4 regulation of pros- overexpressed in LNCaP cells to levels comparable to the endoge- tate xenograft tumor growth and castration resistance in vivo. We nous MAPK4 levels in VCaP and LAPC4 cells (Supplemental Fig- first performed xenograft studies using the LNCaP-iMAPK4 cells ure 2A), both MAPK4D254A and MAPK4S186A mutants retained their in both intact and castrated SCID mice. Dox-induced MAPK4 ability to induce GATA2 and AR expression and AR activation, overexpression greatly enhanced LNCaP xenograft growth in the indicating that MAPK4 regulates GATA2/AR signaling and AKT intact mice, confirming the potent tumor-promoting activity of signaling through distinct mechanisms (Figure 7, B and C). MAPK4 in PCa in vivo (Figure 8A). Furthermore, while only 4 of Since AKT activation is essential for mediating MAPK4 onco- 9 control LNCaP xenografts formed detectable tumors in the cas- genic activity (35), we next compared the biological activities of trated SCID mice 9 weeks after inoculation, MAPK4 overexpres- WT MAPK4 and the MAPK4D254A and MAPK4S186A mutants in pro- sion both increased the tumor incidence to 100% and promoted moting LNCaP cell growth. Interestingly, WT MAPK4, but not the growth of these tumors (Figure 8B). Conversely, Dox-induced MAPK4D254A or MAPK4S186A, robustly promoted LNCaP cell pro- knockdown of MAPK4 greatly inhibited LAPC4 xenograft growth liferation and anchorage-independent growth (Figure 7, D and E). in vivo (Figure 8C). Western blot analysis confirmed the associa- In accord with this, the AKT inhibitor MK2206 greatly inhibited tion of MAPK4 expression with GATA2, AR, and AKT activation in MAPK4-induced LNCaP cell growth (Figure 7F), as well as that of these tumors (Figure 8E). MAPK4-high LAPC4 cells and 22Rv1 cells (Supplemental Figure Finally, to probe how knockdown of MAPK4 affects the 2B). We also confirmed that MK2206 treatment did not repress growth of previously established CRPC tumors, we performed MAPK4-induced GATA2 and AR expression and AR activation in xenograft studies using the 22Rv1-ishMAPK4 cells and the control LNCaP, LAPC4, and 22Rv1 cells (Figure 7F; Supplemental Figure 22Rv1-iNT cells on castrated SCID mice. After xenografts grew 2, C-E). Altogether, these data are in accord with previous results into about 100 mm3, the mice began to receive 4 mg/mL Dox in showing that AKT activation is essential for mediating MAPK4 drinking water to induce knockdown of MAPK4 in the 22Rv1- tumor-promoting activity (35). ishMAPK4 tumors but not in the control 22Rv1-iNT tumors. Since MAPK4D254A and MAPK4S186A mutants were able to Knockdown of MAPK4 again greatly inhibited the growth of these enhance GATA2/AR signaling but failed to robustly promote previously established CRPC xenograft tumors (Figure 8D). Alto- LNCaP cell growth, MAPK4-induced GATA2/AR signaling alone gether, these data support a critical role of MAPK4 in driving PCa appeared not sufficient to mediate the growth-promoting effects growth and castration resistance in vivo. of MAPK4. Activation of AKT alone by overexpression of a consti- AR activation in human PCa tissues is subjected to complex tutively active AKT1T308D/S473D (DD) mutant was also not sufficient regulation, including circulating/intratumor steroid levels, steroid to substantially promote LNCaP cell growth. However, overex- synthesis/metabolism (43–46), AR amplification/overexpression/ pression of this AKT1(DD) mutant largely rescued MAPK4D254A mutation (47–49), AR splice variants (36–41), AR coactivators/

J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 9 RESEARCH ARTICLE The Journal of Clinical Investigation

10 J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 The Journal of Clinical Investigation RESEARCH ARTICLE

Figure 7. Concerted Activation of AKT and GATA2/AR signaling is essen- tance. In accord with this, GATA2 knockdown repressed MAPK4- tial for mediating MAPK4 tumor-promoting activity in PCa. (A) Western induced AR expression/activation, and knockdown of either blots on LAPC4 cells and VCaP cells with knockdown of MAPK4 (G2, G4) GATA2 or AR inhibited MAPK4-mediated PCa cell growth. or control (NT). (B) Western blots, (C) qPCR, (D) Proliferation assays, and (E) Soft-agar assays (original magnification: × 50) on the engineered We previously identified 2 MAPK4 mutants (D254A and S186A) LNCaP cells with 0.5 μg/mL Dox-induced overexpression of MAPK4 (iWT), with greatly decreased ability to activate AKT (35). Interesting- MAPK4S186A (iS186A), MAPK4D254A (iD254A), or control (iCtrl). Data represent ly, these 2 mutants exhibited few growth-promoting effects when mean ± SEM. Adjusted P values determined by 1-way ANOVA followed overexpressed in LNCaP cells. Coexpression of a constitutively by Dunnett’s multiple comparisons. ***P ≤ 0.001. ****P ≤ 0.0001. (F) activated AKT (AKT1T308D/S473D) largely rescued the growth-promot- Proliferation assays and Western blots on the LNCaP cells with 0.5 μg/ mL Dox-induced overexpression of MAPK4 (iWT), MAPK4D254A (iD254A), or ing activities of these 2 mutants, supporting that AKT activation is control (iCtrl), treated with 1 μM of AKT inhibitor MK2206 or DMSO control. essential for mediating MAPK4 activities in PCa. This is consistent (G) Proliferation assays and Western blots on the LNCaP cells with 0.5 μg/ with our previous results on the essential role of AKT in mediating mL Dox-induced expression of MAPK4, MAPK4D254A, or control, also infect- MAPK4 tumor-promoting activities in other types of cancers (35). ed with lentivirus expressing AKT1-DD mutant or control. (H) Proliferation In accord with this, AKT inhibitor treatment blocked the growth assays on the LNCaP cells with 0.5 μg/mL Dox-induced overexpression of MAPK4, MAPK4D254A, GATA2, or control, also overexpressing AKT1-DD or promoting activities of MAPK4 in PCa. These data shed light on control. The growth of these cells in 10% FBS, 5% CSS, and 5% CSS plus 1 the notion that MAPK4-high PCa, especially the MAPK4-high ther- nM R1881 were compared. Also shown are Western blots on these cells cul- apy-resistant CRPC, may benefit from the AKT blockade therapy. tured in 10% FBS. (I) Proliferation assays and Western blots on the control Overexpression of GATA2 alone or activation of AKT alone (NT) or MAPK4-knockdown (G2) VCaP cells with 0.5 μg/mL Dox-induced (overexpression of AKT1T308D/S473D) exhibited a moderate effect on expression of GATA2 (iGATA2), AKT1-DD (iAKT1DD), or control (iCtrl). Data are representative of at least 3 independent experiments. LNCaP cell growth (Figure 7). In contrast, coexpression of both GATA2 and AKT1-DD mutant robustly promoted LNCaP cell growth, which largely recapitulated MAPK4 activity in these cells (Figure 7). These data support the concerted activation of both corepressors (50), protein (31, 51–56), and tumor micro- GATA2/AR and AKT as dual mediators of MAPK4 biology in PCa. environment (57). To enhance our ability to detect the impact The detailed molecular mechanisms underlying the synergistic of MAPK4 on AR activation in human PCa, we specifically ana- activities of GATA2 and AKT in driving PCa remain to be defined. lyzed CRPC from patients in whom circulating steroid levels and MAPK4 activation promotes both hormone-independent and ligand-induced AR activation were reduced. Analysis of the gene hormone-dependent AR transactivation (Figure 4). The augmen- expression profile of laser-captured PCa cells from nonneuroen- tation of agonist response sensitizes PCa cells to subnanomolar docrine CRPC tissues of deceased patients (58) confirmed a posi- androgen levels. Although the mechanisms remain to be deter- tive correlation between MAPK4 expression and the expression of mined, AR overexpression has been suggested to enhance AR AR targets KLK2, PSA/KLK3, the Nelson AR signature (59), and activation in PCa and the strong induction of AR expression in the TCGA AR signature (http://software.broadinstitute.org/gsea/ response to MAPK4 presumably contributes to this potentiation. msigdb/cards/NELSON_RESPONSE_TO_ANDROGEN_UP.html) Importantly, the effect of MAPK4 extends to the clinically signif- (Figure 8F). Altogether, these data support a positive impact of icant AR splicing variant AR-V7, which can drive castration resis- MAPK4 activation on AR in human PCa. tance of PCa (36–41). AKT activation has been indicated in the development of Discussion CRPC, and AR remains activated in most CRPC. MAPK4 can Both AKT and AR signaling pathways play important roles in pro- activate both AKT and AR, suggesting its potent activity in driv- moting PCa. Two landmark studies have revealed that AKT and ing castration resistance of PCa. Our data from LNCaP, LAPC4, AR antagonize each other in PCa (28, 29); repressing one signaling and 22Rv1-based models collectively support this notion. Analysis pathway led to the activation of the other. In contrast, our study further reveals that MAPK4 expression correlates with AR activa- demonstrated that MAPK4 concertedly activates both AR and tion in CRPC tissues (Figure 8F), supporting the human disease AKT and promotes PCa growth and castration resistance. Accord- relevance of our discovery. Therefore, targeting MAPK4 emerges ingly, inhibiting MAPK4 simultaneously represses AR and AKT as a promising therapeutic avenue for a subset of CRPC with sig- activation and inhibits PCa growth, including castration-resistant nificant expression of MAPK4. growth. We believe that these findings support MAPK4 as a novel therapeutic target for PCa, including CRPC. Methods Since the pioneering work from the Brown lab identified GATA2 Reagents and antibodies. Antibodies, including those against HA (cat- as a key factor regulating AR signaling and PCa biology more than alog sc-805), Myc (catalog sc-40), GAPDH (catalog sc-32233), AR a decade ago (6), little progress has been made on understanding (catalog sc-816), Protein A-Agarose (catalog sc-2001), and Protein G molecular mechanisms that regulate GATA2 expression and pro- PLUS-Agarose (catalog sc-2002) were purchased from Santa Cruz tein stability in PCa. Here we demonstrated that MAPK4 enhances Biotechnology. Anti-MAPK4 antibodies used include AP7298b from GATA2 gene transcription through mechanisms that remain to be Abgent, HPA007461 from MilliporeSigma, and TA505872 from Ori- delineated. In addition, MAPK4 promotes GATA2 protein stabili- gene. Anti-GATA2 antibodies used include sc-9008 from Santa Cruz ty through inhibiting its ubiquitination and degradation. By great- Biotechnology and A0677 from ABclonal. Antibodies against p-AKT ly enhancing GATA2 expression/activation, MAPK4 promotes T308 (catalog 4056), p-AKT S473 (catalog 9271), AKT (catalog 2920), AR expression and activation, PCa growth, and castration resis- and p-GSK3β S9 (catalog 9558) were from Cell Signaling Technology.

J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 11 RESEARCH ARTICLE The Journal of Clinical Investigation

Figure 8. MAPK4 promotes prostate tumor growth and its expression correlates with AR activation in human CRPC. Dox-induced overexpression of MAPK4 promotes LNCaP xenograft growth in (A) intact SCID mice and (B) castrated SCID mice. (C) Dox-induced knockdown of MAPK4 inhibits LAPC4 xenograft growth in SCID mice. 2 × 106 of LNCaP-iMAPK4, LNCaP-iCtrl, LAPC4-iNT, or LAPC4-ishMAPK4 cells were injected into the lateral flanks (SubQ) of intact or castrated SCID mice (iCtrl or iNT on the left side; iMAPK4 or ishMAPK4 on the right side). Mice began receiving Dox (0.2 mg/mL for LNCaP and 4 mg/mL for LAPC4 xenografts) in 10% sucrose in drinking water on the day of xenograft implantation. Tumors were harvested as indicated. (D) Dox- induced knockdown of MAPK4 inhibits the growth of previously established 22Rv1 xenograft tumors in castrated SCID mice. A quantity of 2 × 106 22Rv1- iNT or 22Rv1-ishMAPK4 cells were injected into the lateral flanks (SubQ) of castrated SCID mice (iNT on the left side; ishMAPK4 on the right side). Mice began receiving Dox (4 mg/ml) in 10% sucrose in drinking water when tumors grow into about 100 mm3. Tumors were harvested as indicated. iMAPK4 and iCtrl: Dox-inducible expression of MAPK4 or control. ishMAPK4 and iNT: Dox-inducible knockdown of MAPK4 or nontargeting control. Data represent mean ± SEM. P values determined by unpaired 2-tailed Student’s t test. *P ≤ 0.05. **P ≤ 0.01. ***P ≤ 0.001. Data are representative of at least 3 independent experiments. (E) Western blots on an independent set of LAPC4-iNT or LAPC4-ishMAPK4 tumors (a repeated study of panel C). Data are representative of at least 3 independent experiments. (F) MAPK4 mRNA expression correlates with AR activation status in human CRPC tissues.

12 J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 The Journal of Clinical Investigation RESEARCH ARTICLE

EZview Red Anti-FLAG M2 Affinity Gel (catalog F2426), MG132 (cata- Bimolecular fluorescence complementation (BiFC) assay. To direct- log M7449), MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazoli- ly visualize MAPK4 and GATA2 interaction in live cells, we per- um bromide), R1881 (catalog R0908), BrdU (catalog B9285), and anti- formed bimolecular fluorescence complementation (BiFC) as pre- bodies against β-ACTIN (catalog A1978), BrdU (catalog B8434), and viously described (35). In brief, we transfected the YN-MAPK4 and FLAG (catalog F3165) were purchased from MilliporeSigma. Enzalut- YC-GATA2 into HeLa cells and used transfection of YN-MAPK4 and amide (MDV3100, catalog S1250) and MK2206 (catalog S1078) were YC-vector as negative control. At 24 hours after transfection, cells purchased from Selleckchem. were fixed, and the fluorescence signal was detected under micro- Plasmids. The pGIPZ lentiviral shRNA constructs were purchased scope as previously described (35). from Open Biosystem (Thermo Fisher Scientific). The pInducer10 and BrdU assay. Before being seeded on coverslips in 24-well plates, pInducer20 vectors were provided by Thomas Westbrook, Baylor Col- LNCaP-iCtrl/iMAPK4 cells were treated with 0.5 μg/mL Dox in 10% lege of Medicine, Houston, Texas, USA (60). pInducer20-YF vector FBS RPMI-1640 and 22Rv1-iNT/iG2/iG4 cells were treated with 2 was previously generated to allow cDNA subcloning independent of μg/ml Dox in 5% CSS RPMI-1640 plus 10 μM MDV3100 for 3 days. the Gateway system (35). For Dox-inducible knockdown, shRNA oligos The cells were then cultured in presence of 20 μM BrdU for 16 hours were synthesized and cloned into the pLKO-Tet-On vector between and fixed with cold methanol at –20°C for 20 minutes. After treat- AgeI and EcoRI sites. We also used the pInducier10 lentiviral system ment with 2 M HCl for 1 hour, cells were washed with 0.1 M boric acid for the Dox-inducible knockdown of MAPK4, as previously described 3 times and blocked with 1% BSA for 1 hour. Cells were then incu- (35). GATA2 cDNA was PCR amplified from the cDNA prepared from bated with anti-BrdU antibody overnight at 4°C, followed by incuba- the HEK293T cells and cloned into the pRK5 vector (35) between Eco- tion with Dylight 488 goat anti-mouse secondary antibody (catalog RI and SalI sites. GATA2 cDNA was also subcloned into the pInduc- 35502, Invitrogen) for 1 hour. After DAPI counterstaining, images er20-YF vector (35) and the pCDH-CMV-MCS-EF1α-RFP+Puro vector were acquired using a fluorescence microscope. The percentage of (CD516B-2, System Biosciences) to generate the lentiviral vectors for BrdU-positive cells were analyzed and calculated using ImageJ as Dox-inducible and constitutive overexpression of GATA2. The YN and previously described (35). YC vectors for BiFC assay were provided by Xin-Hua Feng, Baylor Col- Quantitative real-time PCR. Total RNA was extracted using the lege of Medicine, Houston, Texas, USA. GATA2 was cloned into the YC Quick-RNA MiniPrep kit (R1054, Zymo Research). Reverse transcrip- vector, and the obtained YC-GATA2 vector was used in the BiFC assay tion reactions were performed using the amfiRivert cDNA Synthesis along with the YN-MAPK4 vector, as previously described (35). Master Mix (R5600, GenDEPOT). Quantitative real-time PCR was per- Cell culture, transfection, and lentivirus infection. LNCaP, LAPC4, formed using the PerfeCTa SYBR Green FastMix reagent (95072, Quan- 22Rv1, and VCaP cells were acquired from ATCC. PNT1A cells were ta Biosciences) on the LightCycler 480 Real-Time PCR System (Roche acquired from the European Collection of Authenticated Cell Cul- Applied Science). PCR primers used are AR: 5′-GCCTTGCTCTCTAG- tures (ECACC). LNCaP and 22Rv1 cells were cultured in RPMI-1640 CCTCAA-3′ and 5′-GTCGTCCACGTGTAAGTTGC-3′, PSA: supplemented with 10% FBS (Hyclone or Invitrogen), and VCaP and 5′-ACCAGAGGAGTTCTTGACCCCAAA-3′ and 5′-CCCCAGAAT- PNT1A cells in DMEM supplemented with 10% FBS. LAPC4 cells CACCCGAGCAG-3′, TMPRSS2: 5′-CGCTGGCCTACTCTGGAA-3′ were cultured in IMDM supplemented with 10% FBS and 1nM R1881. and 5′-CTGAGGAGTCGCACTCTATCC-3′, MAPK4: 5′-TACGGG- Charcoal-stripped FBS (catalog SH3007103) was from Hyclone. GAGAATGCTCTTTG-3′ and 5′-CAGGGATGGTCTCCAGGAT-3′, GenMute siRNA transfection reagent (catalog SL100568) and GATA2: 5′-ATCAAGCCCAAGCGAAGACT-3′ and 5′-CATGGTCAGT- LipoD293 DNA transfection reagent (catalog SL100668, SignaGen GGCCTGTTAAC-3′, GAPDH: 5′-AGCCACATCGCTCAGACAC-3′ Laboratories) were used for siRNA and DNA transfection. For lentivi- and 5′-GCCCAATACGACCAAATCC-3′. Relative expression was ral-mediated gene delivery, lentiviral-based expression constructs were determined using the ΔΔCt method and normalized to GAPDH. transfected into the HEK293T cells together with the packaging mix in Ubiquitination assay. The FLAG-tagged GATA2 cDNA and the Trans-lentiviral packaging kit (catalog TLP5914, Thermo Fisher Sci- HA-tagged ubiquitin in pRK5 vector were cotransfected into HEK293T entific). Viruses in the conditioned media were harvested at 48 hours cells. Forty-eight hours later, cells were treated with 10 μM MG132 for to 96 hours after the transfection and filtered (0.45 μm). The cells were 4–6 hours. The cell lysate was prepared and immunoprecipitation infected with the viral medium for 3 days before drug selection. The was performed using EZview Red Anti-FLAG M2 Affinity Gel (Milli- established stabled cell lines were then expanded for further assays. poreSigma). The IP products were then applied in Western blot assay The pCDH and pInducer20-YF based lentiviral gene delivery sys- using an anti-HA antibody to reveal GATA2 ubiquitination. tem were used for stable and Dox-inducible overexpression of MAPK4, Proliferation assay. For MTT assay, 5 mg/mL of MTT substrate dis- AR, or GATA2 in LNCaP, LAPC4, or 22Rv1 cells. The pGIPZ, pLKO- solved in 1× PBS was added to the cell cultures and incubated at 37°C Tet-On, or pInducer10 vectors were used for lentiviral-mediated sta- for 4 hours. The purple crystals were then dissolved in isopropanol ble or Dox-inducible knockdown of MAPK4, AR, or GATA2 in VCaP, with 0.1% Triton X-100. Absorbance was read at 590 nm. When appli- LAPC4, LNCaP, or 22Rv1 cells. The inducible cells were treated with 2 cable, cells were pretreated with 2 μg/mL Dox for 3 days for inducible μg/mL Dox for at least 3 days to obtain significant MAPK4 knockdown gene knockdown or 0.5 μg/mL Dox for inducible overexpression. Dox or 0.2–0.5 μg/mL Dox for at least 3 days for MAPK4 overexpression. was used throughout the assay. Western blot. Protein samples for Western blot assays were pre- For crystal violet staining–based cell proliferation assay, cells were pared in RIPA buffer. Cell lysate was applied to a brief sonication fixed in 4% paraformaldehyde in PBS for 15 minutes and stained with before protein quantification using the BCA protein assay kit (cata- 0.1% crystal violet in 10% ethanol for 20 minutes. After washing and log 23225, Pierce). An equal amount (5–20 μg) of protein was used in air dry, crystal violet was extracted using 10% acetic acid solution Western blot analysis. (600 μL for 24-well plate). The absorbance was similarly at 590 nm.

J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 13 RESEARCH ARTICLE The Journal of Clinical Investigation

The first measurement was done 24 hours after cell seeding and cell GTTATCTGTCAGA. The following synthesized oligos containing proliferation was assessed every other day as described. G2 and G4 targeting sequences were cloned into the pLKO-Tet- Soft agar colony formation assay. Cells were mixed in full medium On vector between AgeI and EcoRI to generate pLKO-G2 (also containing 0.35% low melting agarose at 37°C and then plated onto named as iG2) and pLKO-G4 (also as iG4). pLKO-G2 (iG2): CCG- supporting gel (1% agarose). When applicable, 2 μg/mL Dox was used GGGGTTGGTAACAAAGTGGTCTCGAGACCACTTTGTTAC- for inducible gene knockdown and 0.5 μg/mL Dox for inducible overex- CAACCCTTTTT (top); AATTAAAAAGGGTTGGTAACAAAGTG­ pression. Colony numbers were counted after inoculation for 3–4 weeks. GTCTCGAGACCACTTTGTTACCAACCC (bottom). pLKO-G4 Xenograft tumor models. Male SCID/beige mice, 8 to 10 weeks old, (iG4): CCGGCGGGAGGAAGACAAGGACGCTCGAGCGTCCTT­ from Envigo were used for xenograft studies. Animals were housed in GTCTTCCTCCCGTTTTT (top); AATTAAAAACGGGAGGAAGA- a pathogen-free facility, and all animal studies were performed under CAAGGACGCTCGAGCGTCCTTGTCTTCCTCCCG (bottom). a Baylor College of Medicine IACUC–approved protocol. For LNCaP Analysis of human tumor molecular data sets. Published microar- xenograft studies, 2 × 106 LNCaP-iMAPK4 or LNCaP-iCtrl cells (pre- ray data (GEO GSE77930) of 171 mCRPC tumors were laser-capture treated with 0.2 μg/mL Dox for 3 days) in 1:1 Matrigel (354234, Corn- micro-dissected, hybridized to Agilent whole-genome microarrays, ing) were injected into the lateral flanks (subQ) of intact or castrated and analyzed as previously described (58). Sample phenotypic groups SCID mice (iCtrl on the left side and iMAPK4 on the right side). Mice were assigned by clustering using classical multidimensional scaling began receiving 0.2 mg/mL Dox in 10% sucrose in drinking water on (MDS) calculated with the cmdscale function in R using the expression the day of tumor inoculation for Dox-induced MAPK4 overexpression profiles of genes from androgen and neuroendocrine signatures (58). in the LNCaP-iMAPK4 xenografts. For LAPC4 xenograft studies, 2 × The profiles of those samples not characterized as neuroendocrine by 106 LAPC4-ishMAPK4 or LAPC4-iNT cells in 1:1 Matrigel were injected transcriptomic signature (n = 144) were further used to analyze gene into the lateral flanks of SCID mice (ishMAPK4 on the left side and iNT expression associations. The signature-normalized enrichment scores on the right side). Mice began receiving 4 mg/mL Dox in 10% sucrose were calculated in R using the GSVA Bioconductor package with in drinking water on the day of tumor inoculation for induced MAPK4 default parameters (61). Pearson correlations of were knockdown in the LAPC4-ishMAPK4 xenografts. For 22Rv1 xenograft calculated by the cor function in R. studies, 2 × 106 22Rv1-ishMAPK4 or 22Rv1-iNT cells in 1:1 Matrigel Statistics. The statistical relevance in the cell culture studies and were injected into the lateral flanks of castrated SCID mice (ishMAPK4 xenograft tumor studies was analyzed using the unpaired 2-tailed Stu- on the left side and iNT on the right side). When tumors grew to about dent’s t test. When there were multiple groups in the analysis, 1-way 100 mm3, mice began receiving 4 mg/mL Dox in 10% sucrose in drink- or 2-way ANOVA followed by Dunnett’s multiple comparisons test ing water for inducible knockdown of MAPK4 in the established 22Rv1- or Sidak’s multiple comparisons test was performed using GraphPad ishMAPK4 xenografts. In all studies, the tumors were monitored every Prism 9. P less than 0.05 was considered significant. 2 days and harvested as indicated and weighed. Tumor size was calcu- Study approval. All animal studies were approved by the Institu- lated as volume = 0.52 × abc (where a, b, and c are the maximum length tional Animal Care and Use Committee of Baylor College of Medicine. of each dimension of the tumor). Average tumor weight was compared between groups for statistical relevance using the unpaired 2-tailed Stu- Author contributions dent’s t test. P less than 0.05 was considered statistically significant. TS, WW, BD, and FY designed the experiments. TS, IC, BD, MMI, siRNAs and shRNAs for knockdown. Synthetic siRNAs targeting DRR, PSN, DDM, and FY wrote and revised the manuscript. TS, GATA2 and AR were obtained from Invitrogen as follows. WW, WZ, QC, YB, and YM performed the experiments. TS, WW, siRNA AR-1: CCGGAAGCUGAAGAAACUUGGUAAU (sense); IC, BD, CJC, and FY analyzed the data. AUUACCAAGUUUCUUCAGCUUCCGG (anti-sense). siRNA AR-2: GAUGAAGCUUCUGGGUGUCACUAUG (sense); CAUAGUG- Acknowledgments ACACCCAGAAGCUUCAUC (anti-sense). siRNA GATA2-1: This research was supported by grants from the Department of CCAGACGAGGUGGACGUCUUCUUCA (sense); UGAAGAAGAC- Defense Congressionally Directed Medical Research Programs GUCCACCUCGUCUGG (anti-sense). siRNA GATA2-2: (W81XWH-13-1-0162, W81XWH-13-1-0163, W81XWH-17-1-0043 GGAAGAUGUCCAACAAGUCCAAGAA (sense); UUCUUGGACUU- to FY) and the Cancer Prevention and Research Institute of Texas GUUGGACAUCUUCC (anti-sense). siRNA against luciferase (siLuc) (RP130651 to FY). We acknowledge the joint participation by Adri- was used as negative control: GCACUCUGAUUGACAAAUACGAUUU enne Helis Malvin Medical Research Foundation through its direct (sense); AAAUCGUAUUUGUCAAUCAGAGUGC (anti-sense). engagement in the continuous active conduct of medical research Lentiviral shRNAs expression constructs were obtained from Open in conjunction with Baylor College of Medicine and the MAPK4 as Biosystem (Thermo Fisher Scientific) with the following targeting a Novel Therapeutic Target for Human Cancers Cancer Program. sequences: pGIPZ-shMAPK4-G2: GGGTTGGTAACAAAGTGGT; pGIPZ-shMAPK4-G4: CGGGAGGAAGACAAGGACG; pGIPZ- Address correspondence to: Feng Yang, Baylor College of Med- shMAPK4-G5: CGGAAGGTCGCTGTGAAGA; pGIPZ-shMAPK4-G6: icine, One Baylor Plaza, Houston, Texas, 77030, USA. Phone: AGGATCGTTGATCAGCATT; pGIPZ-shMAPK4-G7: ACAAGG- 713.798.8022; Email: [email protected].

1. Siegel RL, et al. Cancer statistics, 2019. CA Can- androgen injection on serum phosphatases in vival in metastatic prostate cancer. N Engl J Med. cer J Clin. 2019;69(1):7–34. metastatic carcinoma of the prostate. CA Cancer 2011;364(21):1995–2005. 2. Huggins C, Hodges CV. Studies on prostatic can- J Clin. 1972;22(4):232–240. 4. Scher HI, et al. Increased survival with enzalut- cer. I. The effect of castration, of estrogen and 3. de Bono JS, et al. Abiraterone and increased sur- amide in prostate cancer after chemotherapy.

14 J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 The Journal of Clinical Investigation RESEARCH ARTICLE

N Engl J Med. 2012;367(13):1187–1197. 23. Gewinner C, et al. Evidence that inositol poly- androgen receptor splice variant. J Clin Invest. 5. Smith MR, et al. Apalutamide treatment and phosphate 4-phosphatase type II is a tumor sup- 2010;120(8):2715–2730. metastasis-free survival in prostate cancer. pressor that inhibits PI3K signaling. Cancer Cell. 41. Antonarakis ES, et al. AR-V7 and resistance to N Engl J Med. 2018;378(15):1408–1418. 2009;16(2):115–125. enzalutamide and abiraterone in prostate cancer. 6. Wang Q, et al. A hierarchical network of 24. Klempner SJ, et al. What a tangled web we weave: N Engl J Med. 2014;371(11):1028–1038. transcription factors governs androgen recep- emerging resistance mechanisms to inhibition of 42. Hu CD, et al. Visualization of interactions among tor-dependent prostate cancer growth. Mol Cell. the phosphoinositide 3-kinase pathway. Cancer bZIP and Rel family proteins in living cells using 2007;27(3):380–392. Discov. 2013;3(12):1345–1354. bimolecular fluorescence complementation. Mol 7. Wu D, et al. Three-tiered role of the pioneer fac- 25. Fruman DA, Rommel C. PI3K and cancer: les- Cell. 2002;9(4):789–798. tor GATA2 in promoting androgen-dependent sons, challenges and opportunities. Nat Rev Drug 43. Geller J, et al. Steroid levels in cancer of the gene expression in prostate cancer. Nucleic Acids Discov. 2014;13(2):140–156. prostate—markers of tumor differentiation and Res. 2014;42(6):3607–3622. 26. Taylor BS, et al. Integrative genomic profil- adequacy of anti-androgen therapy. Prog Clin Biol 8. He B, et al. GATA2 facilitates steroid recep- ing of human prostate cancer. Cancer Cell. Res. 1979;33:103–111. tor coactivator recruitment to the androgen 2010;18(1):11–22. 44. Mohler JL, et al. The androgen axis in receptor complex. Proc Natl Acad Sci U S A. 27. Bitting RL, Armstrong AJ. Targeting the recurrent prostate cancer. Clin Cancer Res. 2014;111(51):18261–18266. PI3K/Akt/mTOR pathway in castration- 2004;10(2):440–448. 9. Bohm M, et al. A role for GATA-2 in transition resistant prostate cancer. Endocr Relat Cancer. 45. Montgomery RB, et al. Maintenance of intratu- to an aggressive phenotype in prostate cancer 2013;20(3):R83–R99. moral androgens in metastatic prostate cancer: through modulation of key androgen-regulated 28. Carver BS, et al. Reciprocal feedback regulation a mechanism for castration-resistant tumor genes. Oncogene. 2009;28(43):3847–3856. of PI3K and androgen receptor signaling in growth. Cancer Res. 2008;68(11):4447–4454. 10. Chiang YT, et al. GATA2 as a potential metasta- PTEN-deficient prostate cancer. Cancer Cell. 46. Massie CE, et al. The androgen receptor fuels pros- sis-driving gene in prostate cancer. Oncotarget. 2011;19(5):575–586. tate cancer by regulating central metabolism and 2014;5(2):451–461. 29. Mulholland DJ, et al. Cell autonomous role of biosynthesis. EMBO J. 2011;30(13):2719–2733. 11. Vidal SJ, et al. A targetable GATA2-IGF2 axis PTEN in regulating castration-resistant prostate 47. Visakorpi T, et al. In vivo amplification of confers aggressiveness in lethal prostate cancer. cancer growth. Cancer Cell. 2011;19(6):792–804. the androgen receptor gene and progres- Cancer Cell. 2015;27(2):223–239. 30. Wen Y, et al. HER-2/neu promotes androgen- sion of human prostate cancer. Nat Genet. 12. Minegishi N, et al. Rapid turnover of GATA-2 via independent survival and growth of prostate 1995;9(4):401–406. ubiquitin-proteasome protein degradation path- cancer cells through the Akt pathway. Cancer Res. 48. Chen CD, et al. Molecular determinants of way. Genes Cells. 2005;10(7):693–704. 2000;60(24):6841–6845. resistance to antiandrogen therapy. Nat Med. 13. Lurie LJ, et al. Differential GATA factor stabil- 31. Mellinghoff IK, et al. HER2/neu kinase-depen- 2004;10(1):33–39. ities: implications for chromatin occupancy by dent modulation of androgen receptor function 49. Taplin ME, et al. Androgen receptor mutations in structurally similar transcription factors. Bio- through effects on DNA binding and stability. androgen-independent prostate cancer: cancer chemistry. 2008;47(3):859–869. Cancer Cell. 2004;6(5):517–527. and leukemia Group B study 9663. J Clin Oncol. 14. Whitman M, et al. Association of phosphati- 32. Fang Z, et al. Androgen receptor enhances p27 2003;21(14):2673–2678. dylinositol kinase activity with polyoma mid- degradation in prostate cancer cells through 50. Heemers HV, Tindall DJ. Androgen receptor (AR) dle-T competent for transformation. Nature. rapid and selective TORC2 activation. J Biol coregulators: a diversity of functions converging 1985;315(6016):239–242. Chem. 2012;287(3):2090–2098. on and regulating the AR transcriptional com- 15. Auger KR, et al. PDGF-dependent tyrosine 33. Aberg E, et al. Regulation of MAPK-activated plex. Endocr Rev. 2007;28(7):778–808. phosphorylation stimulates production of novel protein kinase 5 activity and subcellular localiza- 51. Culig Z, et al. Androgen receptor activation polyphosphoinositides in intact cells. Cell. tion by the atypical MAPK ERK4/MAPK4. J Biol in prostatic tumor cell lines by insulin-like 1989;57(1):167–175. Chem. 2006;281(46):35499–35510. growth factor-I, keratinocyte growth factor, 16. Franke TF, et al. The protein kinase encoded 34. Kant S, et al. Characterization of the atypical and epidermal growth factor. Cancer Res. by the Akt proto-oncogene is a target of the MAPK ERK4 and its activation of the MAPK- 1994;54(20):5474–5478. PDGF-activated phosphatidylinositol 3-kinase. activated protein kinase MK5. J Biol Chem. 52. Hobisch A, et al. Interleukin-6 regulates pros- Cell. 1995;81(5):727–736. 2006;281(46):35511–35519. tate-specific protein expression in prostate 17. Sarbassov DD, et al. Phosphorylation and regu- 35. Wang W, et al. MAPK4 overexpression pro- carcinoma cells by activation of the androgen lation of Akt/PKB by the rictor-mTOR complex. motes tumor progression via noncanonical receptor. Cancer Res. 1998;58(20):4640–4645. Science. 2005;307(5712):1098–1101. activation of AKT/mTOR signaling. J Clin Invest. 53. Craft N, et al. A mechanism for hormone-inde- 18. Alessi DR, et al. Characterization of a 3-phospho- 2019;129(3):1015–1029. pendent prostate cancer through modulation of inositide-dependent protein kinase which phos- 36. Dehm SM, et al. Splicing of a novel andro- androgen receptor signaling by the HER-2/neu phorylates and activates protein kinase Balpha. gen receptor exon generates a constitutively tyrosine kinase. Nat Med. 1999;5(3):280–285. Curr Biol. 1997;7(4):261–269. active androgen receptor that mediates pros- 54. Signoretti S, et al. Her-2-neu expression and 19. Zoncu R, et al. mTOR: from growth signal inte- tate cancer therapy resistance. Cancer Res. progression toward androgen independence gration to cancer, diabetes and ageing. Nat Rev 2008;68(13):5469–5477. in human prostate cancer. J Natl Cancer Inst. Mol Cell Biol. 2011;12(1):21–35. 37. Dehm SM, Tindall DJ. Alternatively spliced 2000;92(23):1918–1925. 20. Steck PA, et al. Identification of a candidate androgen receptor variants. Endocr Relat Cancer. 55. Guo Z, et al. Regulation of androgen receptor tumour suppressor gene, MMAC1, at 2011;18(5):R183–R196. activity by tyrosine phosphorylation. Cancer Cell. 10q23.3 that is mutated in multiple advanced can- 38. Guo Z, et al. A novel androgen receptor 2006;10(4):309–319. cers. Nat Genet. 1997;15(4):356–362. splice variant is up-regulated during prostate 56. Mahajan NP, et al. Activated Cdc42-associ- 21. Maehama T, Dixon JE. The tumor suppres- cancer progression and promotes andro- ated kinase Ack1 promotes prostate cancer sor, PTEN/MMAC1, dephosphorylates gen depletion-resistant growth. Cancer Res. progression via androgen receptor tyrosine the lipid second messenger, phosphatidy- 2009;69(6):2305–2313. phosphorylation. Proc Natl Acad Sci U S A. linositol 3,4,5-trisphosphate. J Biol Chem. 39. Hu R, et al. Ligand-independent androgen 2007;104(20):8438–8443. 1998;273(22):13375–13378. receptor variants derived from splicing of cryptic 57. Yang F, et al. Stromal TGF-β signaling induces 22. Gao T, et al. PHLPP: a phosphatase that directly exons signify hormone-refractory prostate can- AR activation in prostate cancer. Oncotarget. dephosphorylates Akt, promotes apopto- cer. Cancer Res. 2009;69(1):16–22. 2014;5(21):10854–10869. sis, and suppresses tumor growth. Mol Cell. 40. Sun S, et al. Castration resistance in human pros- 58. Kumar A, et al. Substantial interindividual and 2005;18(1):13–24. tate cancer is conferred by a frequently occurring limited intraindividual genomic diversity among

J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465 15 RESEARCH ARTICLE The Journal of Clinical Investigation

tumors from men with metastatic prostate can- FGF signaling. Cancer Cell. 2017;32(4):474–489.e6. 2011;108(9):3665–3670. cer. Nat Med. 2016;22(4):369–378. 60. Meerbrey KL, et al. The pINDUCER lenti- 61. Hanzelmann S, et al. GSVA: gene set variation 59. Bluemn EG, et al. Androgen receptor pathway- viral toolkit for inducible RNA interference analysis for microarray and RNA-seqdata. BMC independent prostate cancer is sustained through in vitro and in vivo. Proc Natl Acad Sci U S A. Bioinformatics. 2013;14:7.

16 J Clin Invest. 2021;131(4):e135465 https://doi.org/10.1172/JCI135465